Secondary Logo

Journal Logo

General Articles

The Bispectral Index

A Measure of Depth of Sleep?

Sleigh, James W. MBChB, FANZCA; Andrzejowski, John MBChB, FRCA; Steyn-Ross, Alistair MSc, MNZIP; Steyn-Ross, Moira DPhil, MNZIP

Author Information
doi: 10.1213/00000539-199903000-00035
  • Free


The Bispectral Index (BIS), a measure of depth of anesthesia and sedation [1], is a complex index, but it predominantly quantifies, on a scale of 0 to 100, the degree of coherence among different frequencies in the electroencephalograph (EEG) signal [2]. If the low frequencies are in phase with the high frequencies, the resultant high BIS values reflect good cortical integration: this occurs in the conscious state. Increasing depth of anesthesia results in a decreasing cortical integration and a decreasing BIS score. Typically, BIS values range from 40 to 55 during general anesthesia [1].

Some studies have investigated the changes in BIS that occur during sedation [3,4], but none have investigated how the BIS is affected by natural sleep. Therefore, we describe a small descriptive case series.


After ethical committee approval and written, informed consent, the EEG signal was recorded from five volunteers (aged 14-43 yr) using the Aspect A-1000 EEG monitor (Aspect Medical Systems, Natick, MA). None of the subjects were taking any psychotropic medication.

Commercially available silver-silver chloride EEG pads (Zipprep; Aspect Medical Systems) were attached to the subject's forehead according to a standard montage. One was the ground (Fpz), and the other two were placed over the left and right prefrontal cortex (F7-F8) giving a bifrontal bipolar signal [5]. The low- and high- frequency filters were set to 0.25 Hz and 30 Hz, respectively. The electrode impedances were all <10,000 Omega. We used conventional definitions of the stages of sleep according to EEG patterns [6]:

1. Light sleep (Stages 1 and 2). The patient exhibited a high-frequency/low-amplitude EEG signal, with the presence of sleep spindles or K complexes.

2. Slow-wave sleep (Stages 3 and 4). The EEG displayed high-amplitude/low-frequency waves (manifest as an increase in amplitude in the delta and theta bands of the EEG power spectrum).

3. Rapid eye movement (REM) sleep. In this sleep pattern, there is a low-amplitude/high-frequency EEG, similar to Stage 1 sleep, but with co-existing hypotonia and eyeball movement artifacts.

In addition to visual inspection of EEG waveform for spindles, K complexes and eye movements, the processed data (BIS, 95% spectral-edge frequency [SEF], spectral power in the alpha, beta, delta, theta, and electromyograph [EMG] wavebands) were down-loaded at 10-s intervals onto a computer for subsequent graphing and analysis. We used the EMG power as an estimate of muscle tone. It is well known that there is little EEG spectral power detectable above 50 Hz. The Aspect monitor has therefore defined the power in the frequency range 70-110 Hz as indicative of EMG activity, particularly when frontal electrode placement is used. Assuming that it is not contaminated by external electrical noise, the changes in the power in this waveband probably reflect changes in frontalis muscle activity. Recordings were made for at least one sleep cycle (70-120 min). To facilitate ease of sleep, all the studies were performed in the subject's home environment. They went to bed at 11 PM, and the EEG recording started after a short period of reading to relax. No subjects had received psychoactive drugs or ethanol in the week before the data collection.


There were dramatic changes in the BIS that were remarkably consistent among all five subjects. The changes closely mirrored the classical descriptions of sleep architecture for the early part of the night (Figure 1). After a latency of 10-20 min, the subject descended through light sleep to slow-wave sleep for approximately 60 min. This was followed by a return to light sleep and a short period of REM sleep. The cycle then repeated. Sleep spindles started to seem in the EEG at BIS levels of 75-90. Slow-wave sleep was associated with BIS levels in the range of 20-70. The BIS value dropped very low. The minimal BIS in all cases was <47 and was in the 20s in three subjects. REM sleep occurred with the BIS in the range of 75-92. Each time the subject awoke briefly, the BIS abruptly increased to >90 (usually >96).

Figure 1
Figure 1:
Typical examples of changes in the Bispectral Index during one sleep cycle.

The BIS and 95% SEF both decreased significantly with increased depth of sleep (P < 0.05 t-test with Bonferroni correction for multiple comparisons) (Table 1). Interestingly, unlike the BIS, the SEF did not increase significantly in the transition from slow-wave sleep to REM sleep. This may be due to the artifactual effects of the eye movement on the EEG spectral power. Because the BIS mainly reflects the interfrequency phase-coupling in the EEG signal, it is relatively independent of spectral power and is therefore more resistant to artifacts. The SEF was significantly decreased during REM sleep (P < 0.05).

Table 1
Table 1:
Values of EEG Indices at the Start of Each Sleep Stage


Even our very small study clearly demonstrates that changes in the depth of natural sleep are reflected sensitively by changes in the BIS. Although we have not validated this in large numbers of subjects or against a full 12-lead EEG, the BIS may be a simple indicator of depth of sleep. If so, the Aspect monitor could become a useful component in postoperative somnographic studies. The accuracy of the EMG band output from the Aspect monitor has not been formally validated. Increased sedation or depth of sleep is usually associated with decreased muscle tone, as measured by the submental EMG signal [6]. It is unclear whether our results are due to inaccuracies in the instrumentation or to true differences among responses in different muscle groups. The main deficiency in this study was the lack of validation to a full polysomnogram (12-lead EEG, four-channel electrooculogram, and submental EMG). For this reason, we did not subdivide the sleep stages into four, as is customary, but rather into the coarser gradations of light sleep and slow-wave sleep based on visual inspection of the raw EEG waveform. Because lateral eye movement is well detected by the frontal EEG and there was a noticeable decrease in the EMG power, most REM sleep episodes were probably detected. However, it is conceivable that awakenings and REM sleep could be easily confused. Therefore, more formal validation is necessary before the monitor could be recommended for routine somnography.

Our study also highlighted the EEG similarities of general anesthesia and sleep. The BIS is a measure of the degree to which the low- and high-frequency components of the EEG signal are in phase. The BIS decreases with increased midazolam or propofol sedation from values of approximately 95 when the subject was fully awake, to approximately 70 when subjects became unresponsive to verbal stimulus [3,4]. Levels of propofol necessary to obtund movement in response to a surgical stimulus typically result in BIS values of <50 [1,7]. Because deep sleep causes the BIS to decrease to levels equivalent to a very deep esthetic state, we may conclude that sleep produces, by endogenous mechanisms, a state of gross cortical desynchronization. Our data suggest that the BIS is, to some degree, a nonspecific measure of the level of consciousness, independent of how the loss of consciousness is caused. This has some implications for anesthesiologists, particularly when an anesthetic is being administered using a technique combining regional and general anesthesia (or sedation). Using such a technique, it is not possible to distinguish the effects on the BIS of natural or "endogenous" sedation (due to deafferentation and anxiolysis), from the "exogenous" direct hypnotic effect of sedative drugs on the BIS value. The raw BIS value may not reliably quantify how easily the patient may be aroused. Minor intraoperative stimuli may waken a sleeping patient from a very low BIS value if they have a predominantly endogenous component to their sedation.

We conclude that similarity in BIS measurements indicate that the transition from consciousness to the natural sleeping state is similar to the transition to unconsciousness caused by anesthesia.


1. Vernon JM, Long E, Sebel PS, Manberg P. Prediction of movement using bispectral electroencephalographic analysis during propofol/alfentanil or isoflurane/alfentanil anesthesia. Anesth Analg 1995;80:780-5.
2. Sigl JC, Chamoun NC. An introduction to bispectral analysis for the electroencephalogram. J Clin Monit 1994;10:392-404.
3. Liu J, Singh H, White PF. Electroencephalogram bispectral analysis predicts the depth of midazolam-induced sedation. Anesthesiology 1996;84:64-9.
4. Kearse LA, Rosow C, Slavsky A, et al. Bispectral analysis of the EEG predicts conscious processing of information during propofol sedation hypnosis. Anesthesiology 1998;88:25-34.
5. Martin JH. The collective electrical behaviour of cortical neurons. In: Kandel ER, Schwartz JH, Jessel TM, eds. Principles of neural science. 3rd ed. London: Prentice Hall, 1991:779.
6. Culebras A. Clinical handbook of sleep disorders. Boston: Butterworth-Heinemann, 1996:107-9.
7. Leslie K, Sessler DI, Smith WD, et al. Prediction of movement during propofol/nitrous oxide anesthesia. Anesthesiology 1996;84:52-63.
© 1999 International Anesthesia Research Society